Reconfigurable microstrip antenna optimization through artificial neural networks

Abstract

Artificial neural networks are applied to reconfigurable microstrip antenna design and optimization in this paper. The change of resistor value leads to two different results, exploited as the less and the high-performance models and align them as a solution through artificial neural networks. Two phases of optimization are employed by (I) using multilayer perceptron, and (II) prior knowledge input methods. The optimization techniques are implemented through Matlab. In the proposed antenna example, the S-parameter is optimized in separate optimization processes. The proposed reconfigurable microstrip antenna resonates initially at dual-band between 1-5 GHz. The antenna is modified by adding a control circuit included separately two resistors and switches between both conductor parts. By increasing the value of resistors the decrease in the return loss at optimal operating frequencies is observed. © 2020 IEEE.